Precoder Design and Power Allocation for Downlink MIMO-NOMA via Simultaneous Triangularization

TL;DR

This paper introduces a novel precoding scheme for downlink MIMO-NOMA systems that improves ergodic rate performance by avoiding channel inversion, and develops a power allocation algorithm to optimize the achievable rate region.

Contribution

The paper proposes a simultaneous triangularization-based precoding scheme for MIMO-NOMA that outperforms existing diagonalization methods and approaches the dirty paper coding limit.

Findings

The proposed scheme avoids channel inversion, enhancing ergodic rates.

It outperforms baseline SD-based precoding and orthogonal access schemes.

Hybrid time-sharing further improves the rate region.

Abstract

In this paper, we consider the downlink precoder design for two-user power-domain multiple-input multiple-output (MIMO) non-orthogonal multiple access (NOMA) systems. The proposed precoding scheme is based on simultaneous triangularization and decomposes the MIMO-NOMA channels of the two users into multiple single-input single-output NOMA channels, assuming low-complexity self-interference cancellation at the users. In contrast to the precoding schemes based on simultaneous diagonalization (SD), the proposed scheme avoids inverting the MIMO channels of the users, thereby enhancing the ergodic rate performance. Furthermore, we develop a power allocation algorithm based on the convex-concave procedure, and exploit it to obtain the ergodic achievable rate region of the proposed MIMO-NOMA scheme. Our results illustrate that the proposed scheme outperforms baseline precoding schemes based on SD and orthogonal multiple access for a wide range of user rates and performs close to the dirty paper coding upper bound. The ergodic rate region can further be improved by utilizing a hybrid scheme based on time sharing between the proposed MIMO-NOMA scheme and point-to-point MIMO.